Daily Governance Digest

China's AI companion rules took effect today, shutting down Doubao and Qwen persona features for hundreds of millions; the EU AI Act's August 2 GPAI enforcement date is 18 days away with retroactive fines possible; China's MOFCOM export restriction consultations continue with no final decision.

China's AI companion rules take effect tomorrow as ByteDance completes Doubao shutdowns; the FTC opens comment period on AI accuracy policy through July 31; and the EU AI Act's August 2 GPAI enforcement date is now 19 days away.

ByteDance and Alibaba are actively shutting down AI companion features ahead of China's July 15 enforcement deadline; the EU Digital Omnibus still awaits Official Journal publication as August 2 Article 50 enforcement nears; and Beijing's MOFCOM consultations on restricting overseas access to advanced Chinese AI models remain unresolved.

Apple files federal lawsuit against OpenAI over alleged AI hardware trade secret theft; European Commission releases EU Action Plan on Cybersecurity and AI; California lawmakers leave 30+ AI bills in appropriations limbo as the legislature breaks for summer recess.

The first UN Global Dialogue on AI Governance closes with a call for concrete action before 2027; China's AI companion and agent rules take effect in four days; and the EU Digital Omnibus awaits Official Journal publication as August 2 enforcement approaches.

China's Ministry of Commerce holds discussions with Alibaba, ByteDance, and Z.ai on potential overseas AI model access restrictions; EU AI Act transparency code signatory deadline is July 22; House lawmakers release first comprehensive bipartisan federal AI governance bill draft.

GPT-5.6 Sol, Terra, and Luna launch publicly today completing the first government-supervised pre-release cycle under EO 14409; the ESRB warns frontier AI models pose systemic cyber risk to EU financial markets.

US Commerce Department cleared OpenAI's GPT-5.6 model family for global release on July 9, ending the first government-supervised pre-release cycle under EO 14409. EU GPAI enforcement activates in 25 days on August 2 as the Digital Omnibus still awaits Official Journal publication.

Illinois Governor Pritzker signed a landmark AI safety law on July 6 joining California and New York to cover roughly 40% of the US AI market; the FTC AI accuracy policy statement was formally published in the Federal Register today, opening the 30-day comment period.

The UN's first intergovernmental AI governance dialogue opened in Geneva today with warnings that AI is outpacing human control; the EU AI Act Digital Omnibus awaits Official Journal publication as the August 2 GPAI enforcement deadline is 27 days away; and Anthropic has deployed technical controls to block Chinese firms accessing Claude through offshore workarounds.

OpenAI floats giving the US government a 5% equity stake in an AI wealth fund; the UN convenes its first intergovernmental AI governance dialogue in Geneva; and the UK Data Act rewrites automated decision-making rules in a divergence from the EU approach.

The FTC proposed July 1 that secretly suppressing AI output accuracy for undisclosed ideological reasons could violate consumer protection law. Court emails in the Anthropic-Pentagon dispute show the Defense Department conditioned contracts on accepting autonomous weapons use, while Anthropic moves to close offshore loopholes exploited by Chinese firms.

China's five-agency anthropomorphic AI rules take effect in 12 days, the EU AI Act Omnibus nears Official Journal publication before August 2, and Tennessee's AI mental health advertising ban is now in force.

Anthropic's Fable 5 returns to global availability after the US lifts its export-control directive, ending a 19-day shutdown; the FTC opens comment on an AI accuracy policy statement; and Congress unveils a bipartisan 269-page federal AI framework discussion draft.

Vermont's neurological rights and AI law takes effect today making it the fifth state with neural data protections; the Trump AI executive order's cybersecurity clearinghouse deadline expires tomorrow; and the EU AI Office's binding enforcement powers over GPAI models activate in 32 days.

Colorado's original AI Act reaches its original effective date having been replaced by a narrower ADMT law; the EU finalizes a timeline extension for high-risk AI compliance; Rhode Island enacts three AI laws targeting mental health chatbots and healthcare AI.

Trump signs executive order directing a frontier AI model framework and cybersecurity clearinghouse; Congress releases the first comprehensive federal AI bill draft; OpenAI restricts new model access to trusted partners at the U.S. government's request.